services/sambanova_voice_service.py

import io
import base64
import json
import os
from pathlib import Path
import gradio as gr
import numpy as np
import openai
from fastrtc import (
    AdditionalOutputs,
    ReplyOnStopWords,
    Stream,
    get_stt_model,
    get_twilio_turn_credentials,
)

class SambanovaVoiceService:
    """Dịch vụ Sambanova AI với Voice Streaming hoàn chỉnh"""
    
    def __init__(self, tts_service=None):
        # Khởi tạo client Sambanova
        self.client = openai.OpenAI(
            api_key=os.environ.get("SAMBANOVA_API_KEY"),
            base_url="https://api.sambanova.ai/v1",
        )
        # Khởi tạo STT model
        self.stt_model = get_stt_model()
        # Tích hợp TTS service
        self.tts_service = tts_service
        print("✅ Sambanova Voice Service initialized với TTS")
    
    def get_available_models(self):
        """Lấy danh sách model có sẵn"""
        return [
            "Meta-Llama-3.1-8B-Instruct",
            "Meta-Llama-3.1-70B-Instruct"
        ]
    
    def generate_response(self, messages, model="Meta-Llama-3.1-8B-Instruct", temperature=0.1, top_p=0.1):
        """Generate response từ Sambanova API"""
        try:
            response = self.client.chat.completions.create(
                model=model,
                messages=messages,
                temperature=temperature,
                top_p=top_p,
                max_tokens=1024,
            )
            return response.choices[0].message.content
        except Exception as e:
            print(f"❌ Sambanova API Error: {e}")
            return f"Xin lỗi, có lỗi xảy ra: {str(e)}"
    
    def stream_generate_response(self, messages, model="Meta-Llama-3.1-8B-Instruct", temperature=0.1, top_p=0.1):
        """Stream response từ Sambanova API"""
        try:
            response = self.client.chat.completions.create(
                model=model,
                messages=messages,
                temperature=temperature,
                top_p=top_p,
                max_tokens=1024,
                stream=True
            )
            
            full_response = ""
            for chunk in response:
                if chunk.choices[0].delta.content:
                    text_chunk = chunk.choices[0].delta.content
                    full_response += text_chunk
                    yield text_chunk, full_response
                    
        except Exception as e:
            error_msg = f"❌ Lỗi: {str(e)}"
            yield error_msg, error_msg
    
    def speech_to_text(self, audio):
        """Chuyển speech thành text"""
        try:
            text = self.stt_model.stt(audio)
            print(f"🎤 STT Result: {text}")
            return text
        except Exception as e:
            print(f"❌ STT Error: {e}")
            return ""
    
    def text_to_speech(self, text, language='vi'):
        """Chuyển text thành speech sử dụng TTS service"""
        if self.tts_service is None:
            print("❌ TTS service chưa được khởi tạo")
            return None
        
        try:
            audio_bytes = self.tts_service.text_to_speech(text, language)
            if audio_bytes:
                # Lưu file tạm và trả về filepath
                filename = f"tts_{int(time.time())}.mp3"
                filepath = self.tts_service.save_tts_audio(audio_bytes, filename)
                return filepath
            return None
        except Exception as e:
            print(f"❌ TTS Error: {e}")
            return None
    
    def generate_response_with_voice(self, messages, model="Meta-Llama-3.1-8B-Instruct", language='vi'):
        """Generate response và chuyển thành voice"""
        try:
            # Lấy text response
            text_response = self.generate_response(messages, model)
            
            # Chuyển thành voice
            audio_filepath = self.text_to_speech(text_response, language)
            
            return {
                "text": text_response,
                "audio": audio_filepath,
                "audio_bytes": self.tts_service.text_to_speech(text_response, language) if audio_filepath else None
            }
        except Exception as e:
            print(f"❌ Error in generate_response_with_voice: {e}")
            return {
                "text": f"Xin lỗi, có lỗi xảy ra: {str(e)}",
                "audio": None,
                "audio_bytes": None
            }